Titre : Orientation and Connectivity Based Criteria for Asymptotic Consensus
Exposant : Bernadette Charron-Bost
Résumé : We consider a set of autonomous agents that interact with each other by
exchanging values and perform instantaneous operations on received values.
Agents each start with a real value and must reach agreement on a value which
is a convex combination of the initial values. The agents are not required to agree
exactly, but ought to iteratively compute values that all converge to the same limit.
For such multiagent systems, the asymptotic consensus problem has been proposed
a solution which consists in an iterative linear procedure, classically referred to as the
"agreement algorithm". It has been originally introduced for the synchronous and time-
invariant case, and then has been extended to the case of asynchronous communications
and time-varying environment. A related algorithm has been later proposed as a model
of a natural algorithm for cooperative behavior. The subject has recently attracted considerable
interest within the context of flocking and multiagent coordination.
In this work, we establish orientation and connectivity based criteria for the agreement
algorithm to achieve asymptotic consensus in the context of time-varying topology and
communication delays. These criteria unify and extend many earlier convergence results
on the agreement algorithm for deterministic and discrete-time multiagent systems.